Test A.4.1, in which parsing depends on reconstructed palette index values.
Test A.4.2, in which parsing is independent of reconstructed palette index values.
Some gain was shown – 0 to 2% per source content category (all-intra).
The results for A.4.1 are not significantly better than for A.4.2.
The average number of context coded bins is increased, but not the worst case. Two additional contexts are used.
Test A.5 – Contexts for run coding
Proponent: Qualcomm, JCTVC-R0174 part 1, JCTVC-S0038
Crosschecker: Canon, JCTVC-S0071
In this test, use of additional contexts depending on the index for coding index runs is tested. Additional contexts for greater than 0, greater than 1 and greater than 2 flags will be tested. The context is dependent on the index value (decoded) or the coded index (parsed). Additionally choice of context based on index value and the palette size was tested.
Dostları ilə paylaş: |